Virtual tours provide a mechanism for a user to view different parts of a physical location or site from different perspectives and in different directions. A virtual camera moves from room to room, turns to present the view in different directions from a particular point in space, and performs other operations that allow a user to engage in a virtual tour. In some cases, users can control the movement, orientation, and focal length of the virtual camera.
Such virtual tours are useful in many different contexts where it may be advantageous for a user to see a physical site from different angles and points of view. Examples of physical sites that can be presented in a virtual tour include:                a house or other property for sale;        a hotel room or cruise ship stateroom;        a museum, art gallery, or other sightseeing destination;        a nuclear plant for training purposes;        a street view.        
Virtual tours are often provided over the World Wide Web, via a website. Such web-based virtual tours may be presented via any suitable format, such as for example Adobe Flash.
Virtual tours are often created by stitching together a number of digital photographs. A digital camera may be rotated to various positions, so that images can be captured in various orientations. In some cases, specialized equipment may be used, such as a fish eye lens to enable floor-to-ceiling views of the physical location. In either case, the creation of the virtual tour requires that a specific set of images be captured, and that the captured images have a defined spatial relationship with one another that can be exploited in creating the virtual tour.
However, even with a complete set of digital photographs covering a location or scene, most existing technologies are unable to provide an immersive three-dimensional virtual tour. To the contrary, in general such tours can be generated only if specialized equipment is used to create a three-dimensional model of the environment; such equipment may be costly, difficult to use, and not readily available to an ordinary user. Accordingly, high-quality virtual tours are, in general, only possible for physical locations where such equipment can be installed and used. Without the ability to capture a three-dimensional model of the physical location, conventional methods for generating virtual tours fail to provide a mechanism for generating satisfactorily immersive tours.
Furthermore, existing techniques for providing virtual tours generally consume large amounts of bandwidth, since significant amounts of data must be transmitted in order to provide a satisfactory user experience.
Furthermore, it is often difficult to make changes to a virtual tour, since changing one aspect or portion of the tour generally requires re-rendering the entire tour.
Furthermore, existing techniques for providing virtual tours generally require that a user navigate through a scene by dragging a mouse or pressing keys on a keyboard. Transitions from one image to the next take place in response to user action. Conventional mechanisms generally provide no technique by which transitions from one image to the next take place automatically without requiring user input.
Furthermore, many existing techniques for providing virtual tours introduce discontinuities when transitioning from one image to the next, which can cause the user to lose his or her spatial reference.